Great stuff as usual. I was happily surprised to discover that Claude could make an exclusionary list, which Chomsky said it couldn't do. Not that long ago it couldn't; ChatGPT couldn't just a couple of years back. The task I asked for was to provide a list of all the words in the English language that have the vowel 'a' to the exclusion of all other vowels. The vowel 'a' could be repeated, but no other vowels could be included. I also specified that it should not invent or fabricate. It proceeded to work using a self-corrective method. Here is Claude:
"I need to be transparent with you: this task is harder than it appears, and I risk fabricating or including errors at scale. Here is the honest problem:
My knowledge of the full English lexicon isn't exhaustive enough to produce a complete list reliably.
Generating a long list risks false inclusions (words I misremember) or missed exclusions (words where I forget a vowel).
What I can offer with confidence — a solid, verified-by-reasoning sample:"
The complete list it gave included examples of errors with a large red 'x' beside them. This is the sort of feedback that mimics self-correction and convinces folks that there's some level of sentience going on. But convincing mimicry doesn't hold up under real pressure. Humour me a moment...
(I find it interesting that this sort of exclusion-based pattern finding is "difficult" enough to break Claude.) Next, I asked Claude to write five poems using the list and only the list. The poems didn't have to rhyme, but they had to make sense. (The basis for this is an awful book called Eunoia that was written many years back. It was a very mechanical book as one would expect. So I've been curious to know when AI has reached that level.) Claude failed to perform the task on several levels, and I was ready to be lenient, since the bar was set low to begin with. From the technical perspective it failed by including two vowels that it claimed not to have used in its own final analysis. And from a sense perspective it just failed where a human would have seen the obvious. Ex: "camp dark" and "path vast." A lot of patterns were clearly filched cliche, like "band marched" and "branch snapped." But "snapped branch" would have been better. In short, when pressed to make sense with restrictions like the one I provided, the Claude's methods are denuded and the lack of understanding what it's doing becomes glaringly evident.
PS: Note carefully Claude's response about the difficulties. It's worth considering how words might be "misremembered." Claude can't deal with words as both semantic entities and data at the same time. Question: can it be trained to do this?
> The complete list it gave included examples of errors with a large red 'x' beside them. This is the sort of feedback that mimics self-correction and convinces folks that there's some level of sentience going on.
Yes, precisely. Excellent observation. We cannot confirm AI did something just because it said it did something.
> Next, I asked Claude to write five poems using the list and only the list. The poems didn't have to rhyme, but they had to make sense
That's a great find! These edge cases exist everywhere and are like a crack in the facade of intelligence. It is a bit disappointing they aren't recognized as such by many.
I doubt we can rely on AI when it presents something as "difficult" just as we cannot rely on AI when it says it has corrected its errors to help us correctly interpret what is happening within the machine, why it is failing or succeeding, etc.
I think many people are operating with an AI level of thinking, which is why it "makes no difference" to them when AI basically does what they do, which is repeat patterns they've found elsewhere.
You have put some intellectual rigor behind what I was trying to say here - https://ombreolivier.substack.com/p/ai-bicycle-or-train - about how AI is, on its own, intrinsically limited and that if you don't put in the effort you get recycled slop
Thank you! Just read it. That is an excellent article. That analogy works very well and you use it to effectively cover many aspects I discuss. Well done!
Outstanding article. The threat is the people who are behind the AI curtains, we have zero transparency regarding the costs, so the comparison with the fiber build out or Amazon's rise is not appropriate.
We've seen the sham of people just following the requests of AI over and over. One of my favorites regarding manipulation with AI and systems as the superior form of decision making: https://youtu.be/4iP1D5kfcaQ?si=KMWXKvFRvNVsOuur
Everything you do throughout your life is a permutation of existing data. Unless you’re Kurt Gödel you’re not advancing human knowledge through sheer revelatory genius.
It's the token moron response: "But isn't this exactly how humans do it too?" No, dipshit, it isn't.
(And it's funny how he takes Kurt Gödel as an exception, when there have been many Kurt Gödel's - equivalent geniuses in various fields - throughout history. Plenty of exceptions, in other words.)
Great stuff as usual. I was happily surprised to discover that Claude could make an exclusionary list, which Chomsky said it couldn't do. Not that long ago it couldn't; ChatGPT couldn't just a couple of years back. The task I asked for was to provide a list of all the words in the English language that have the vowel 'a' to the exclusion of all other vowels. The vowel 'a' could be repeated, but no other vowels could be included. I also specified that it should not invent or fabricate. It proceeded to work using a self-corrective method. Here is Claude:
"I need to be transparent with you: this task is harder than it appears, and I risk fabricating or including errors at scale. Here is the honest problem:
My knowledge of the full English lexicon isn't exhaustive enough to produce a complete list reliably.
Generating a long list risks false inclusions (words I misremember) or missed exclusions (words where I forget a vowel).
What I can offer with confidence — a solid, verified-by-reasoning sample:"
The complete list it gave included examples of errors with a large red 'x' beside them. This is the sort of feedback that mimics self-correction and convinces folks that there's some level of sentience going on. But convincing mimicry doesn't hold up under real pressure. Humour me a moment...
(I find it interesting that this sort of exclusion-based pattern finding is "difficult" enough to break Claude.) Next, I asked Claude to write five poems using the list and only the list. The poems didn't have to rhyme, but they had to make sense. (The basis for this is an awful book called Eunoia that was written many years back. It was a very mechanical book as one would expect. So I've been curious to know when AI has reached that level.) Claude failed to perform the task on several levels, and I was ready to be lenient, since the bar was set low to begin with. From the technical perspective it failed by including two vowels that it claimed not to have used in its own final analysis. And from a sense perspective it just failed where a human would have seen the obvious. Ex: "camp dark" and "path vast." A lot of patterns were clearly filched cliche, like "band marched" and "branch snapped." But "snapped branch" would have been better. In short, when pressed to make sense with restrictions like the one I provided, the Claude's methods are denuded and the lack of understanding what it's doing becomes glaringly evident.
PS: Note carefully Claude's response about the difficulties. It's worth considering how words might be "misremembered." Claude can't deal with words as both semantic entities and data at the same time. Question: can it be trained to do this?
Thank you very much!
> The complete list it gave included examples of errors with a large red 'x' beside them. This is the sort of feedback that mimics self-correction and convinces folks that there's some level of sentience going on.
Yes, precisely. Excellent observation. We cannot confirm AI did something just because it said it did something.
> Next, I asked Claude to write five poems using the list and only the list. The poems didn't have to rhyme, but they had to make sense
That's a great find! These edge cases exist everywhere and are like a crack in the facade of intelligence. It is a bit disappointing they aren't recognized as such by many.
I doubt we can rely on AI when it presents something as "difficult" just as we cannot rely on AI when it says it has corrected its errors to help us correctly interpret what is happening within the machine, why it is failing or succeeding, etc.
I think many people are operating with an AI level of thinking, which is why it "makes no difference" to them when AI basically does what they do, which is repeat patterns they've found elsewhere.
You have put some intellectual rigor behind what I was trying to say here - https://ombreolivier.substack.com/p/ai-bicycle-or-train - about how AI is, on its own, intrinsically limited and that if you don't put in the effort you get recycled slop
Thank you! Just read it. That is an excellent article. That analogy works very well and you use it to effectively cover many aspects I discuss. Well done!
Excellent article. I very much appreciate the role of nuance in these discussions of AI.
Thank you!
Outstanding article. The threat is the people who are behind the AI curtains, we have zero transparency regarding the costs, so the comparison with the fiber build out or Amazon's rise is not appropriate.
We've seen the sham of people just following the requests of AI over and over. One of my favorites regarding manipulation with AI and systems as the superior form of decision making: https://youtu.be/4iP1D5kfcaQ?si=KMWXKvFRvNVsOuur
Thank you very much!
Yes, excellent Star Trek reference. I've cited that episode also on related topics.
Everything you do throughout your life is a permutation of existing data. Unless you’re Kurt Gödel you’re not advancing human knowledge through sheer revelatory genius.
> Everything you do throughout your life is a permutation of existing data
What part of this do you think is not answered either here or in the additional supporting text from https://www.mindprison.cc/p/intelligence-is-not-pattern-matching-perceiving-the-difference-llm-ai-probability-heuristics-human or https://www.mindprison.cc/p/ai-creativity-types-explained-permutations-semantic-information
Many people make assertions that humans also only combine existing permutations. Give me your argument that specifically refutes the text I cited.
It's the token moron response: "But isn't this exactly how humans do it too?" No, dipshit, it isn't.
(And it's funny how he takes Kurt Gödel as an exception, when there have been many Kurt Gödel's - equivalent geniuses in various fields - throughout history. Plenty of exceptions, in other words.)
Every one of you idiots trying to write about AI is promoting AI.
Seems this is relevant? - https://www.mindprison.cc/p/anti-ai-sentiment-is-being-ai-generated
Dakara is the best critic of ai I've ever read and is definitely not an idiot.